Publically accessible, illegal, websites represent an additional challenge for control agencies, but also an opportunity for researchers to monitor, in real time, changes in criminal careers. Using a repeated measures design, we examine evolution in the networks that form around child exploitation (CE) websites, over a period of 60 weeks, and determine which criminal career dimensions predict website failure. Network data were collected using a custom-designed web-crawler. Baseline survival rates were compared to networks surrounding (legal) sexuality and sports websites. Websites containing CE material were no more likely to fail than comparisons. Cox regression analyses suggest that increased volumes of CE code words and images are associated with premature failure. Websites that are more popular have higher odds of survival. We show that traditional criminal career dimensions can be transferred to the context of online CE and constitute some of the key determinants of an interrupted career.